Related Experiment Videos
"Nonparametric" A' and other modern misconceptions about signal detection theory
Richard E Pastore1, Edward J Crawley, Melody S Berens
1Department of Psychology, Binghamton University, SUNY, Binghamton, New York 13902-6000, USA. pastore@binghamton.edu
Psychonomic Bulletin & Review
|November 19, 2003
Summary
Signal detection theory (SDT) is often misunderstood, leading to flawed cognitive process conclusions. This study clarifies SDT's origins and assumptions, contrasting them with modern distortions to improve research validity.
Area of Science:
- Cognitive psychology
- Psychophysics
- Decision science
Background:
- Modern Signal Detection Theory (SDT) descriptions often misrepresent the Gaussian equal-variance model (G-SDT).
- These distortions can lead to incorrect conclusions about cognitive processes.
- Some researchers avoid standard SDT measures (d', beta) due to assumption concerns, opting for less reliable alternatives (A', B").
Purpose of the Study:
- To clarify the origins, development, and assumptions of SDT.
- To contrast accurate SDT with common modern distortions and misconceptions.
- To provide guidance on the appropriate use and interpretation of SDT measures.
Main Methods:
- Historical review of Signal Detection Theory.
- Comparative analysis of G-SDT with alternative measures (A', B").
- Examination of statistical assumptions underlying SDT measures.
Main Results:
- Many current SDT applications are distorted caricatures of the original G-SDT model.
- Alternative measures like A' and B" are often misused and confound ability with bias.
- Misinterpretations of SDT can invalidate conclusions about cognitive processes.
Conclusions:
- A clear understanding of SDT's foundations is crucial for accurate research.
- Researchers should critically evaluate their use of SDT measures and the conclusions drawn by others.
- Correct application of SDT enhances the evaluation of cognitive processes and research findings.